(this is a post moved from my old blog. My old blog post now links to this one)
I've been asked this a few times now: "If I have a functional or technical requirement, why do I need tasks? The requirement IS my task, right?"
The answer is, as is the case for most topics of debate in the world of PM/RM/CM/SCM (enough already!), it depends!
(NOTE: Before reading further, try to dissassociate these terms from actual physical entities in a tool. By this I mean, don't think of MS Project whan I mention a "task", and don't think of a Word document when I discuss "requirements.")
Requirements and tasks, while they both can be interpreted as instructions for work, inherently have attributes unique to themselves. To explore and identify these attributes, let's look at the definition of each.
Requirements
A requirement, defined by our friends at dictionary.com is:
Does this definition mention that this is an instruction for work? Yes and no. While a requirement doesn't necessarily state "Go do this," it does have that implication. For example:
FR 101: The system shall provide a search mechanism.
All vagueness aside, this requirement doesn't actually directly say, "Build a search mechanism". It more simply states that in order for this project to be successful, there needs to be a search mechanism. But the task itself is implied: "The system shall provide a search mechanism[, so go make one]."
Requirements are success criteria, answering "Why are we doing this?", and "What exactly is this thing going to be?"
If you have a relatively small team, detailed requirements, and don't see the need to record work done against requirements implementation in much detail, then simply using requirements as "assignments of work" may be perfectly feasible.
Tasks
A task is defined as:
Tasks, by definition, are much more explicit in their direction to a particular consumer. They specifically attribute themselves to a consumer as a unit of work to be done.
Tasks are work criteria, answering "What exactly am I supposed to do?"
If you have a need for more granular work tracking, or your requirements are defined in a way that it may take multiple consumers or efforts to implement them, I'd recommend using tasks to at least some degree.
Using Requirements and Tasks Together
The easiest way to see a marriage between requirements and tasks is by admitting that some requirements may need multiple things to be done to be implemented. Take our earlier sample requirement:
Simple enough, right? Well, to implement this requirement, perhaps a search dialog needs to be created. A database query needs to be established. The actual search operation itself may need to be threaded to minimize impact to the foreground application. If you're a PM managing your development team, you may want to know more specifically what your developer is working on today than, "I worked 5 hours on FR 101". You will see better reporting metrics if instead you can see, "I worked 2 hours on the search dialog, and 3 hours on the database query."
So what? Why does all this matter? Whether or not you use both requirements and tasks, or just one type of artifact, is ultimately up to you. But take a gander at user scenarios/needs that I've experienced, and you might get a better idea of what to do in your environment:
Well, that's my 2 cents. I know there are many opinions out there (I'm sure there are even more approaches!) about this topic. The beauty of having multiple solutions to an issue such as is this that it allows us PM's and SCM guys to keep jobs!
Two scenarios will be discussed in this post: Single Hat and Multiple Hats.
Base Scenario:
You have 3 people: Joe, Sally, and Dave
You have 3 main roles: Developer, Tester, Reviewer
You also have 3 projects: Project A, Project B, and Project C
Scenario #1: Single Hats
Team members wear only one hat in the enterprise. A Developer for one project is a developer for all projects – the same for Tester and Reviewer.
The roles that Joe, Sally and Dave play are the same for every project:
Developer
Tester
Reviewer
Project A
Joe
Sally
Dave
Project B
Project C
The simple setup for this in Team Foundation Server is to use generic role-based groups:
Team Foundation Server
\Developers
\Joe
\Testers
\Sally
\Reviewers
\Dave
When configuring your Team Project’s permissions, simply grant each group the desired rights. This will allow any subsequent users to be added to the environment with ease (just add them to the group that fits their role).
Scenario #2: Multiple Hats
Your team may have roles that vary by project. A good way to support this in Team Foundation Server is to create role-based groups on a per-project basis.
The roles that Joe, Sally and Dave play vary with each project:
The inherent problem with using generic role-based groups (as in Scenario #1) is that in this scenario, everyone would have full rights to each of the three projects because each person belongs to each group:
A more practical approach is to use project-specific, role-specific groups. This adds several extra groups, but more effectively manages access control at the project level:
\Project A - Developers
\Project A - Testers
\Project A - Reviewers
\Project B - Developers
\Project B - Testers
\Project B - Reviewers
\Project C - Developers
\Project C - Testers
\ Project C - Reviewers
I was recently asked about how date/time stamps are stored in the TFS. The concern was centered around distributed development that crosses timezones. If someone in the UK checks in a piece of code to a server in California, how will the stamp be preserved so that a California user knows precisely when the code was modified?
Gone are the days of the 500 user recommendation:
http://blogs.msdn.com/bharry/archive/2005/12/09/502190.aspx